Keithley 182 leakage

[gamalot]

It took me some time to get familiar with chart.js and Flask and finally made a simple logging application.

Then I did a test. The meter was disconnected from the 182 in the first 3 minutes, and the zero offset was about 1.7pA, I compensated in the software. When the meter is connected to 182, the current is too large, in the chart I limit it to 30pA to avoid scaling problems.

It can be seen that the leakage current is about 4pA after 10 minutes, and the voltage of the battery is 8.8 volts.

[Kleinstein]

Much of what one sees as leakage can be dielectric absorbtion in the isolation material. So the current is expected to go down quite a bit.  Som of the isolation materials (PCB board and PVC) have low conductivity, but still rather high dielectric losses.

One should also be able to see the other side of dielectric absorbtion: first apply a voltage between ground and the 182 for quite some time and after that measure the current with no more applied voltage.

The impedance between the isolated DMM part and ground is probably well decribes as a rather lossy capacitor, not so much a resistor in the TOhms. It is still a bit tricky to tell appart low levels of true leakage current and dielectric absorbtion. Mainly patience helps there, but it can take hours.

[gamalot]

Much of what one sees as leakage can be dielectric absorbtion in the isolation material. So the current is expected to go down quite a bit.  Som of the isolation materials (PCB board and PVC) have low conductivity, but still rather high dielectric losses.

One should also be able to see the other side of dielectric absorbtion: first apply a voltage between ground and the 182 for quite some time and after that measure the current with no more applied voltage.

The impedance between the isolated DMM part and ground is probably well decribes as a rather lossy capacitor, not so much a resistor in the TOhms. It is still a bit tricky to tell appart low levels of true leakage current and dielectric absorbtion. Mainly patience helps there, but it can take hours.

It does seem to be exactly the same as the charging process of the capacitor. I did another 1 hour test last night, and the result was completely expected, the second half of the curve is flat and can't tell the difference between it and a straight line.

The difference is that the final current is 7pA instead of the previous 4pA, I'm guessing because of changes in temperature and humidity, I'm having the worst winter since moving to Melbourne 9 years ago.

Before using this DIY voltmeter, I forgot to do a gain error test on it, so I did it last night. Applying a 200mVpp/10mHz sine wave voltage to its input through a 1GOhm resistor in series, the result is acceptable.

[gamalot]

I have planned to disassemble the DIY meter into parts and reuse them, but when I went to the supermarket yesterday, I found that there are 12V batteries for sale, so let it work again!

Two MN21 batteries in series is 25.3V, and the leakage current down to 21.4pA after one hour, which is consistent with the previously measured result of 7pA at 8.8V.

[gamalot]

Sorry I'm going off topic again!

When I got the new Amphenol 97-3057-1008-1, I just realized that the 97-3057-1008(621) I ordered years ago was wrong, not just the color, but also missing the bushing. 

[gamalot]

Finally found the Advantest R8240 in the corner of the closet in my room, the voltage source was the same battery pack (25.3V) as last time, and the leakage current was less than 11pA after an hour.

---

I left them powered on for over 10 hours and the temperature in the room is now about 10 degrees hotter than when I measured in the afternoon and the leakage current reading has risen to about 16pA

[guenthert]

Hello Philipp

No 182 here but 2 Keithley 181

Your setup with an external PSU at 10V and 10 MOhm DMM in series does not reveal any usable data for me. The measurements are all over the place. Maybe I have too much noise today in my lab.

So, I used a Keithley 617 Electrometer with the internal 10V output of the 617.

Keithley 181 #1: 26 to 28 pA
Keithley 181 #2: 5 to 6 pA

These readings are taken after about 10 min stabilization time and me being 3 m away.

   The 617 is very slow and will average the current (not sure, how accurate it'll be with varying currents).  The bias current of the 181 (like most DMM) is very peaky though.  Note that the service manual wants one to adjust the pump out current compensation so that the voltage peaks across a 10MOhm resistor [of your oscilloscope's probe] on the input is less than 2mV.  That'll be peak currents of 200pA. 

Edit:
    The service manual for the Keithley 182 has the procedure for verifying that the input bias current meets specification p 7-21 ff.  There indeed the Keithley 617 is used, albeit as coulomb meter.  If that was obvious to everyone but me, then I apology for the noise.